Académique Documents
Professionnel Documents
Culture Documents
and Opportunities,
Editors Hans Müller-Steinhagen, M. Reza Malayeri, and A. Paul Watkinson, Engineering Conferences International, Tomar, Portugal, July 1 - 6, 2007
Reboiler Design Practice that no matter how rigorous design methods are,
there’s simply little evidence that their
The term ‘reboiler’ is generally used in the predictions are reliable. Obviously, before
Process Industry to describe equipment whose releasing design methods to end-users
function is to add a specified duty to a correlations are generally tested with available
distillation or other separation process. They are test data to show that they are effective and
broadly classified as once-through, reliable. However, the two-phase flow with or
thermosyphon, or forced flow. In practice, without heat transfer in industrial heat
however, only three types of reboilers are exchangers, specifically on the shell side, cannot
common: be reproduced with bench-scale test units.
Often, engineers are left to propose a design
• Kettle; based at least in part on engineering judgments
• Vertical thermosyphon or forced and previous experiences, or even an educated
circulation; and guess. The primary objective of the design is to
• Horizontal thermosyphon or forced meet the performance in terms of duty for non-
circulation. fouled conditions. The end users are generally
responsible for taking corrective measures to
The need for careful design of reboilers is well minimize the effects of the two-phase pattern
known. In fact, if operational problems are to be and localized thermal conditions on fouling.
avoided the design process of this type of This current stop-gap practice is generally
equipment calls for very careful thought. responsible for fouling of reboilers and costly
corrective measures the end users have to
Some basic guidelines are: implement in after-the-fact scenario.
• Use a proven thermal design tool(s). Increasingly, designers seeking to provide end
• Check the performance at every possible users with the best possible solution are relying
operating condition, especially at on what is known as evidence-based
maximum turndown and at start-up (no engineering, a hard, cold, empirical look at what
fouling). works what doesn’t work, and how to
• Do not specify an excessive fouling distinguish between the two. This approach has
resistance; this can lead to film boiling been used with varying degree of success, but it
when the exchanger is clean and worse should be effective in dismissing the most
performance than if it were fouled. cherished beliefs in the industry, like the idea
• Always check the flow regimes, that higher fouling factors would result in longer
particularly at maximum turndown. cleaning cycles.
Separated and mist flow regimes should
be avoided. But is this possible or even desirable? Heat
• Consider designing reboilers with low transfer engineering, after all is a service built
residence time, which can be achieved around uniqueness of each particular design and
with low liquid hold up and minimizing engineers’ ability to design equipment
recirculation zones. accordingly. Furthermore, how do we create the
evidence the end-users demand, unless we test
Nobody pretends design of reboilers is the untested? Whatever the merits of “evidence
straightforward, but there is one thing we ought based engineering,” further progress will require
to be able to rely on: it’s that the engineers an increased reliance on “expert” opinion
working on a design should be doing more than supported by knowledge-based prediction
playing hunches. We use certain methods methods.
because they work, right? We use certain
procedures, embedded in commercial software
suites because they work well, don’t we?
However, more often than not engineers realize
http://dc.engconfintl.org/heatexchanger2007/11
Panchal and Ljubicic: 73
Feed/Effluent (F/E) Exchangers single phase flow can be applied to two phase
flows. In order to avoid poor vapor and liquid
F/E exchangers are used to recover process heat phase distribution, vertical welded plate heat
from reactor (hydrotreater and reformer) exchangers are being considered. They show
products and preheat the feed stream, which low fouling propensity; however, the long-term
consists of petroleum feed and recycled performance of such exchangers under varying
hydrogen. In practice the following types of F/E process conditions needs to be validated.
exchangers are used.
The current design practices focus on achieving
• Horizontal shell and tube with U- the thermal performance for a given pressure
tube bundles; drop limit. There are no adequate design
• Vertical combined feed exchanger guidelines to evaluate fouling propensity for
(VCFE) also known as vertical two-phase flow conditions under varying
Texas Tower F/E; and process parameters. As a result, no corrective
• Vertical welded plate exchangers. measures are built into designing F/E
exchangers. Therefore, there is a need to
The two-phase flow feed is normally on tube examine the current practice of designing F/E
side and the product stream is on the shell side. exchangers, develop validated predictive
Non-uniform distribution of the petroleum feed methods for localized fouling/coking conditions,
and recycled hydrogen among tubes of and evaluate new design configurations
horizontal as well as vertical F/E exchangers including alternate heat exchangers.
induces high localized wall temperatures. If
these localized wall temperature are above Major Fouling Mechanisms, Root Cause
threshold fouling conditions at prevailing liquid- Analysis and Mitigation
phase wall shear stress, then formation of
precursors (e.g. decomposition of organo- The three major fouling mechanisms associated
metallic complexes) would occur. Tubes with with reboilers and F/E heat exchangers are:
high vapor flows could produce undesired flow
regimes (such as mist flows) with intermittent 1. Reaction fouling – polymerization;
dryout or thin film, which could produce high 2. Deposition of corrosion products; and
fouling/coking. Tubes with low vapor flows 3. Coking under dryout and/or under
would produce slug or separated flows with high critical heat-flux conditions.
residence time and low wall shear stress would
increase the precursor formation in the liquid Polymerization fouling, generally induced by
phase. Amelse et al [2004] analyzed the iron-sulfide formation in refining, is commonly
distribution of vapor and liquid phase in F/E observed on feed side, generally tube side, of
exchanger of p-xylene plant using radioactive F/E exchangers of hydrotreating and reforming
tracers. They correlated high rates of fouling process units [Limke, 1999]. Polymerization of
with the hydrogen and hydrocarbon distribution. diolefins [Brons et al. 1999] is a common
It was shown that the fouling rate was problem in reboilers and reactor overhead
significantly high in tubes with high liquid to condensers of hydrotreaters, hydrocrackers, and
hydrogen flow ratios. Such conditions produce FCC fractionators. The root cause analysis
low wall shear in the liquid phase, separated should focus on: 1) two-phase flow patterns at
flows, and higher wall temperatures particularly the inlet, particularly on how vapor and liquid
in regions of intermittent vapor-liquid interfaces are fed to the exchanger, 2) flow regimes in
on the wall. All these parameters are return passes, 3) chemical analysis of deposits
responsible for accelerated fouling rates. collected from different parts of the exchanger to
However, correlations are not available to determine the main precursor(s) and hence
determine threshold limits of these parameters identify the most probable chemical mechanism,
for two phase flows in tubes and we are not sure 4) thermal profiles, 5) distribution of fouling
how threshold conditions determined with the deposits within the exchangers, and 6) change in
pressure drop between cleaning. Maldistribution (twisted tubes or spiral baffles) should be
of liquid and vapor phases at the feed point and considered.
also in return passes of multi-pass exchangers
can produce conditions with high propensity of Polymerization fouling may lead to further
localized fouling, which can then propagate to coking reaction within the deposits as wall
the rest of exchangers. Understanding the temperatures increase after significant fouling
governing fouling mechanisms and designing occurs. More commonly coking in reboilers and
the exchanger to operate below threshold wall F/E exchangers is caused by dryout and critical
conditions (temperature and wall shear stress) heat flux with high wall temperatures.
provides necessary design information for an Intermittent wetting and dryout produce droplets
alternate tube bundle to minimize fouling. The and thin film and the transient heat transfer
primary choices for alternate tube bundles are induces coke-like deposits separating out as
different tube size, twisted tubes, and spiral solid or semi-solid phase. The top region of
baffles to alter thermal profiles and two-phase reboilers is prone to coking, if mist flow occurs
flow patterns. If alternate tube bundle cannot and high-temperature heating media (e.g. high-
meet necessary conditions to minimize pressure steam) is used. The root-cause analysis
conditions then consider altogether different heat should focus on analysis of deposits to
exchanger. Exchangers that provide good determine coking mechanisms (solubility tests)
distributions of liquid and vapor phases and and identifying regions with high coking within
minimize the overall resident time would the exchanger. Improved or alternate heat
minimize fouling by reducing the rate of exchanger design to minimize mist flow,
precursor formation in the exchanger. Welded avoiding super heated steam, and evaluating
plate heat exchangers with narrow flow passages threshold coking conditions during turn-down
have been shown to reduce fouling; however, it operation (e.g. use of recirculation pump) are
is important to evaluate individual applications possible mitigation steps based on design.
for suitability of such exchangers. F/E
exchangers with two-phase flow on shell side Application of Single-Phase Fouling Data to
provide low residence time and should have Two-Phase Flows
relatively uniform distributions to minimize
fouling. Fouling data are generally obtained with single-
phase flow fouling units. Such fouling data or
The deposition of corrosion products is more an empirical correlation based on these data
common in reboilers, where corrosion products cannot be directly applied to exchangers with
may be generated in the distillation column or two-phase flows. If the flow regimes are known
overhead condensers/separation drums. These and reliable correlations are available to predict
corrosion products tend to deposit in the local wall shear stress and heat transfer rates,
recirculation zones. The root cause analysis then Kuru and Panchal [1997] showed how the
should focus on deposit analysis and tracking the fouling data obtained with single-phase fouling
source of corrosion. The localized deposition unit could be applied to tube side annular two-
can alter the flow patterns and thermal phase flow regimes.
conditions locally, producing conditions of
higher fouling propensity for organic fouling Panchal et al [1997] developed a threshold
(polymerization) in other regions. Combined fouling model applicable to preheat train. The
corrosion and polymerization fouling can be rate of fouling is expressed in terms of
difficult to manage. However, controlling generation of precursor in the boundary layer
corrosion fouling could minimize organic and its removal before firm adhesion of deposits
fouling. The reboiler in general should have on the wall occurs.
provision purge corrosion products collected at
the feed point. Alternate heat exchanger designs dRf/dt= α Re -β Pr -0.33 exp( -E/RTf) - λ τs (1)
which provide good flow patterns on shell side
http://dc.engconfintl.org/heatexchanger2007/11
Panchal and Ljubicic: 75
Where Reynolds number, Re and Prandtl the heat fluxes and hence TMTs are
number, Pr are calculated for properties at the incorporated, the localized coking rates
bulk and film temperature Tf (average of bulk increases significantly. The distribution of
and metal wall temperatures), respectively. The vapor and liquid phases in the U-bend regions
Reynolds number in this equation was based on further increases propensity of localized coking.
the two-phase flow viscosity. The wall shear Figure 3 shows the CFD-based prediction of
stress, τs is calculated for the liquid flow at the distribution of vapor and liquid phase
wall surface. Constants α, β, λ and the downstream of U-bend. Similar flow
activation energy E are determined from distribution is expected for horizontal F/E
laboratory and/or field data. When the equation exchanges with U-bend tube bundles.
is equated to zero, it provides threshold 2
fouling/coking conditions between in-side tube 1.8 Pipe#1 Pipe#2 Pipe#29 Pipe#30
metal temperature and the wall shear stress. 1.6 Pipe#13 Pipe#14 Pipe#25 Pipe#26
1.4
1.2
The model was developed for single-phase
Coking rate
1
flows; however, it can be applicable to two 0.8
0
preheat train to the process side of a crude fired 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
demonstrate applicability of the fouling model to Fig.2 Coking rate (hr ft2 F/Btu (10-4) per day)
an industrial fired heater to evaluate the distribution.
localized fouling propensity. Figure 1 shows
circumferentially averaged tube-metal
temperature (TMT) for selected pipes of a crude
heater predicted by the process model and API
530 (American Petroleum Institute design
standard 530), which is commonly used by the
petroleum industry to estimate TMTs.
840
820 Pipe#1 Process Model Pipe#1 API530
800
Pipe#30 Process Model Pipe#30 API530
780
760
740
TMT, F
720
700
680
660
640
620
600
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Axial distance from top
The localized coking rate predicted using Eqn. 1 Fig. 3 Cross section at up flow exit of a down
is shown in Figure 2. The predicted results flow U-bend
show gradual increase in the coking rates as the
TMT temperature rises, although the wall shear Figure 3 shows secondary flow patterns and
stress also increases with increase of the vapor accumulation of liquid (dark region) on the outer
quality. When the circumferential variations of
http://dc.engconfintl.org/heatexchanger2007/11
Panchal and Ljubicic: 77
bundle.
The future improvements in both increased As we did not have all operating data required to
energy efficiency and extended runtimes will calculate heat transfer coefficients on a daily
likely come from the use of closer temperature basis, reboiler performance and fouling was
approaches. As the temperature difference assessed or ‘trended’ using indirect process
http://dc.engconfintl.org/heatexchanger2007/11
Panchal and Ljubicic: 79
¾ The uncertainty of the appropriate value Mitigation, Begell House Publication, New
for the critical flux and onset of fouling York, NY. 273-281.
for horizontal F/E exchangers grows as Panchal CB, Lottes S, and Petrick M,
the size of the unit becomes larger. 2006, “Two-phase flow pipe-side process model
There is limited amount of data on for fired heater,” Proceedings of the AIChE
critical flux with geometries and National Meeting, Orlando, FL.
conditions relevant large industrial
designs.
¾ As energy and efficiency of thermal
systems become more important factors
in the overall economics of process
industry, the broader acceptance of
enhancement technologies in industrial
practice will take place.
References